Gone Without a Bang: an Archival HST Survey for Disappearing Massive Stars
Total Page：16
File Type：pdf, Size：1020Kb
MNRAS 453, 2885–2900 (2015) doi:10.1093/mnras/stv1809 Gone without a bang: an archival HST survey for disappearing massive stars Thomas M. Reynolds,1,2‹ Morgan Fraser1 and Gerard Gilmore1 1Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, UK 2Tuorla Observatory, Department of Physics and Astronomy, University of Turku, Vais¨ al¨ antie¨ 20, FI-21500 Piikkio,¨ Finland Downloaded from Accepted 2015 August 4. Received 2015 July 16; in original form 2015 May 27 ABSTRACT It has been argued that a substantial fraction of massive stars may end their lives without http://mnras.oxfordjournals.org/ an optically bright supernova (SN), but rather collapse to form a black hole. Such an event would not be detected by current SN surveys, which are focused on finding bright transients. Kochanek et al. proposed a novel survey for such events, using repeated observations of nearby galaxies to search for the disappearance of a massive star. We present such a survey, using the first systematic analysis of archival Hubble Space Telescope images of nearby galaxies with the aim of identifying evolved massive stars which have disappeared, without an accompanying optically bright SN. We consider a sample of 15 galaxies, with at least three epochs of Hubble Space Telescope imaging taken between 1994 and 2013. Within this data, we find one candidate at CERN - European Organization for Nuclear Research on September 15, 2016 which is consistent with a 25–30 M yellow supergiant which has undergone an optically dark core-collapse. Key words: stars: evolution – stars: massive – supernovae: general. plosion, and Type IIL (Linear) SNe which have a steady decline 1 INTRODUCTION in luminosity from peak. These differing SN types have long been Massive (>8M) stars end their lives as core-collapse supernovae believed to result from the extent of the H and/or He envelope of (CCSNe). Once they have evolved off the main sequence and de- the progenitor star at the moment of collapse (Falk & Arnett 1977), veloped a Chandrasekhar-mass Fe core, the pressure in their core is although recently there has been some debate in the literature as to no longer sufficient to support the star against its own gravity. The whether Type IIP and IIL SNe are distinct classes (e.g. Anderson inner core begins to infall, a process which is halted once it reaches et al. 2014; Faran et al. 2014, and references therein). nuclear densities and forms a protoneutron star (PNS). At this point Nearby CCSNe present a unique opportunity to test this hy- the core rebounds and drives a shock outwards through the star. pothesis by directly identifying their massive stellar progenitors in According to current simulations, this shock lacks the energy to archival images. This was first accomplished for SN 1987A in the halt the stars collapse and explode the outer layers of the star. It is Large Magellanic Cloud (LMC; West et al. 1987), and for SN 1993J believed that the deposition of additional energy behind the shock in M51 (Aldering, Humphreys & Richmond 1994; Maund & Smartt (perhaps in the form of neutrinos) can revive it and cause the star to 2009). Since then, there has been significant success in detecting the explode as a CCSNe (e.g. Janka et al. 2007; Burrows 2013). If the H-rich progenitors of Type II SNe (e.g. Li et al. 2007; Mattila et al. energy deposition is insufficient, then the shock is not revived and 2008; Maund et al. 2011; Van Dyk et al. 2012;Fraseretal.2014). In instead the outer layers of the star collapse on to the PNS, forming a particular, Type IIP SNe have been shown to come from red super- black hole (e.g. Woosley 1993;Fryer1999; O’Connor & Ott 2011). giants (RSGs) with extended atmospheres (Smartt et al. 2009). The This is a more likely outcome for more massive stars, which have H-poor progenitors of Type Ibc SNe, however, have thus far eluded bigger Fe cores and are consequently harder to explode. confirmed detections (Eldridge et al. 2013), with only iPTF13bvn SNe are distinguished primarily by the elements seen in their (Cao et al. 2013; Bersten et al. 2014; Eldridge et al. 2015)asa spectra – Type I SNe are H poor, while Type II SNe are H rich viable, albeit unconfirmed, candidate. (Filippenko 1997;Smartt2009). The Type I SNe are further sub- While statistical studies of ensembles of SN progenitors (Van divided into Type Ib and Type Ic SNe which show the presence Dyk, Li & Filippenko 2003;Smarttetal.2009) have revealed that or absence of He, respectively. H-rich Type II SNe are separated Type IIP SNe do indeed come from RSGs, there is an apparent lack according to their light curves into Type IIP (Plateau) SNe, which of progenitors with a luminosity comparable to that of the brightest show an extended period of roughly constant luminosity after ex- known RSGs. Smartt et al. found that there was a statistically signif- icant absence of higher mass (16 M) RSGs exploding as SNe, terming this the ‘Red Supergiant Problem’. Smartt et al. went on to E-mail: thmire@utu.fi suggest that the RSG problem might be explained by failed SNe, C 2015 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society 2886 T. M. Reynolds, M. Fraser and G. Gilmore withstarsinthemassrangefrom16to30M collapsing to form detected by these surveys, and hence its progenitor will not be iden- black holes with very weak (or perhaps non-existent) explosions. tified (although we note that Yang & Lunardini 2011 have raised Furthermore, attempts have been made to constrain SN progenitor the possibility of detecting failed SNe through their neutrino emis- masses without direct progenitor observations. By analysing the sion). As an alternative approach, Kochanek et al. (2008) proposed resolved stellar population in the vicinity of nearby SNe, the local a ‘survey about nothing’ to monitor a sample of nearby galaxies stellar age can be determined, and used to set an upper mass limit with sufficient depth and resolution to detect individual massive on the SN progenitor (Williams et al. 2014). Williams et al. set 11 stars. By searching for a massive star which had disappeared, it additional (to those of Smartt et al. 2009) progenitor constraints with would be possible to identify faint or dark failed supernovae, which this method with results consistent with there being no progenitors show no bright optical display. Kochanek et al. (2008) suggested with 20 M. either using the Hubble Space Telescope (HST) or ground-based There is some supporting evidence for the suggestion that 8-m class telescopes for such a survey. Gerke, Kochanek & Stanek 16 M stars do not explode as optically bright SNe. Spectra (2015) presented the first data from such a survey, using a sam- Downloaded from of SNe taken ∼1 yr after explosion can provide a diagnostic of ple of 27 galaxies, observed over 4 years with the Large Binocular the core mass and composition of the star that exploded (Mazzali Telescope. Gerke et al. found a single candidate in their data, which et al. 2010; Jerkstrand et al. 2012). In particular, the ejected mass they suggested was consistent with a failed SN fraction of ∼0.3 of of O and the strength of the [O I] emission lines increase rapidly all core-collapses. for a progenitor with a mass of more than ∼20 M (Jerkstrand In this paper, we present the first attempt to conduct a systematic http://mnras.oxfordjournals.org/ et al. 2014). To date, however, there has been no observed Type IIP search for failed supernovae using extant archival HST data.InSec- SNe with strong [O I] emission consistent with a progenitor with a tion 2, we discuss the selection criteria for our sample of galaxies, zero-age main sequence (ZAMS) mass 16 M (Jerkstrand et al. and the technique used to search for disappearing stars; in Section 3, 2014). We also note that Brown & Woosley (2013) found that galac- we present the candidate failed SNe found, while in Sections 4 and 5 tic abundances can be reproduced even if no stars above 25 M we provide a discussion of the results and conclusions, respectively. contribute metals via SN explosions, and that even an even lower threshold of 18 M could be accommodated with some modifica- tion of uncertain mass-loss parameters and reaction rates. 2 SAMPLE AND METHODOLOGY Attempts have also been made to compare the observed SN rate The observational signature of an optically dark core-collapse SN is with the star formation rate (SFR; Horiuchi et al. 2011; Botticella a massive star which is seen to disappear without an accompanying at CERN - European Organization for Nuclear Research on September 15, 2016 et al. 2012). Horiuchi et al. claimed that SN rates were lower than bright SN. Our basic approach is to take galaxies which have been expected given the observed SFRs in galaxies locally and at in- observed with the HST in the same filter, on at least three separate termediate redshifts, and suggested that this may be explained by occasions, and search for any luminous point source which is present a significant number of optically dark (either intrinsically or due in the first two epochs, but is no longer present in the third epoch. to dust obscuration) SNe. However, measurements of both the SN In the following section, we elaborate on our methodology, and rate and SFR suffer from significant systematic and random un- describe in detail the construction of our sample of galaxies and the certainties (Mathews et al. 2014), and Botticella et al. (2012)find selection of data used. that there is no disagreement between the SN rate and SFR within 11 Mpc.